臺灣博碩士論文加值系統

本研究旨於探討人工智能分析社群媒體大數據之旅遊目的地意象，是否在官方文宣發布前後發生變化，進而推論官方文宣之影響力。本研究以人工智能加大數據研究方式，分析2017~2021遊客在日月潭地區主題標籤(Hashtag)「日月潭」之相片，觀察在遊客歷年旅遊目的地意象變化上是否有不同。本研究工具為人工智能Yolov3，研究對象為日月潭官方發布新八景與舊八景，經由訓練人工智能進行影像辨識，下載了共109155張被遊客主題標籤(Hashtag)日月潭的照片進行影像辨識分析。其結果顯示：(1)官方文宣對新八景旅遊意象提升無顯著關係；(2)新八景旅遊意象最高月份為4月，與日月潭光觀總人次沒有關連性；(3)大數據分析出的新八景意象排行與官方顯示排行差距甚大。

The purpose of this research is using artificial intelligence to analyze the big data of tourist destinations image in social media. Whether images changes before and after the official announcement, and then infer the influence of the official announcement. This research uses artificial intelligence to enhance the data research method, to analyze the photos of tourists in the Sun Moon Lake Hashtag "日月潭" from 2017 to 2021, and to observes whether there are differences in the image changes of the tourists' travel destinations over the years. The research tool is artificial intelligence Yolov3 and is trained to recognition images. The research object is Sun Moon Lake's official promotion spots of the New Eight Scenes and the Old Eight Scenes. A total of 109,155 photos of Sun Moon Lake with a tourist hashtag is downloaded for image identification and analysis. The analytic results show that: (1) Official promotion has no significant relationship with the tourism image of the New Eight Scenes. (2) The highest month of the tourism images of the New Eight Scenes is April but it has no correlation with the total number of Sun Moon Lake visitors (3) There is a big gap between the ranking of New Eight Scenes from official promotion and big data.

摘要......i
Abstract......ii
誌謝......iii
目錄......iv
表目錄......vii
圖目錄......ix
壹、緒論......1
1.1研究背景與動機......1
1.2研究目的......2
1.3研究限制......2
1.3.1研究對象......2
1.3.2隱私條款......2
1.3.3 Yolov3......2
1.3.4日月潭舊八景......2
1.4名詞解釋......3
1.5研究流程......4
貳、文獻回顧......5
2.1 Web2.0、使用者生成內容UGC與大數據之關係......5
2.1.1 Web2.0以後誕生的UGC......5
2.1.2 Web2.0的UCG形成大數據......5
2.2社群媒體、旅遊生成內容（TGC）與旅遊目的地意象(TDI)之關係......6
2.2.1 旅遊生成的UGC—TGC......6
2.2.2 旅遊目的地意象......7
2.2.3 TGC與旅遊目的地意象之關係......8
2.3旅遊文宣(NTO，DMO)、社群媒體(TGC)旅遊目的地意象之關係......9
2.3.1 國家旅遊組織NTO、目的地營銷組織DMO......9
2.3.2社群媒體Instagram的影響力......10
2.3.3 TGC與TDI之關係......11
2.3.4 NTO、DMO的管理困境......12
2.4人工智能與旅遊休閒的未來......13
2.4.1人工智能簡介......13
2.4.2未來旅遊與人工智能......14
2.4.3 AI電腦視覺研究上的優勢......14
2.5 YOLOV3與電腦視覺......16
2.5.1 YOLOV3是一種電腦視覺......16
2.5.2 YOLOV3及其研究應用......16
2.5.3硬體上準備......18
2.5.4軟體上準備......18
2.5.5訓練集與測試集準備......18
2.5.6圖像的標籤(Label)......19
2.5.7 YoloV3圖片收集準則與規範......20
2.5.8 YOLO的訓練學習原理......23
2.5.9本章文獻結論歸納......26
2.7本研究相關研究......27
參、研究方法......30
3.1研究步驟與架構圖......30
3.2研究假設......32
3.3研究對象......32
3.4研究工具：人工智能的訓練、測試......35
3.4.1新八景訓練集三階段收集方法......36
3.4.2各景點訓練圖片收集詳細細節......39
3.4.3人工智能模型測試......73
3.4.4以人工實際觀測YoloV3辨識結果......82
3.5研究資料分析方法......89
肆、結果與討論......91
4.1研究大數據相片基本屬性......91
4.2 官方新八景文宣發布後對日月潭旅遊目的地意象之影響力......95
4.2.1 新八景2017~2020卡方分析結果與討論......95
4.2.2 舊八景2017~2020卡方分析結果與討論......100
4.2.3 新八景/舊八景2018~2019卡方分析結果與討論......102
4.2.4 特殊新八景卡方分析結果與討論......105
4.3日月潭新八景各月份意象變化過程......107
4.4日月潭新八景意象排行......119
伍、結論與建議......125
5.1研究結果討論與建議......125
5.1.1圖片大數據基本屬性......125
5.1.2 官方文宣影響力之分析......126
5.1.3新八景各月份意象之變化......128
5.1.4官方文宣與實際分析結果比對---以新八景排行為例......129
5.2研究貢獻......129
5.2.1學術貢獻......129
5.2.2實務貢獻(一)......130
5.2.3實務貢獻(二)......131
5.3研究限制......131
5.3.1 樣本因素......132
5.3.2 氣候疫情因素......132
5.4後續研究建議......132
5.5未來展望......133
參考文獻......135
Extended Abstract......139

中文文獻：
[1]Amazon Machine Learning(2021)。模型擬合：低度擬合與過度擬合。網址：https://docs.aws.amazon.com/zh_tw/machine-learning/latest/dg/model-fit-underfitting-vs-overfitting.html
[2]TWNIC(2020)。2019台灣網路報告。網址：https://report.twnic.tw/2020/
[3]日月潭內政部觀光局(2018)。新8景票選活動簡章。網址：https://files.bountyhunter.co/dl/contest/201807/c4deca06-e50a-4021-ba1e-46adbe5000cb.pdf?dl=2018%E6%97%A5%E6%9C%88%E6%BD%AD%E5%85%AB%E6%99%AF%E7%A5%A8%E9%81%B8%E8%BE%A6%E6%B3%95%E7%B0%A1%E7%AB%A0.pdf
[4]李瑋倫、謝秉融、洪作緒、李沅銘，丁信修、陳大麟(2019)。人工智能與大數據在油氣探採之應用。石油季刊，55(4)，55-69。
[5]林健巍(2019)。YOLO圖像檢測技術綜述。福建電腦，35(9)，80-83。
[6]邱皓政(2014)。量化研究與統計分析：SPSS(PASW)資料分析範例(第五版)。台中市：五南圖書。
[7]南投縣政府(2021)。106~110南投縣觀光遊憩區遊客人次統計表。網址：https://saao.nantou.gov.tw/funcation/tenders_messagesr.aspx?path=157&id=497
[8]施輝、陳先橋、楊英(2019)。改進YOLO v3的安全帽佩戴檢測方法。計算機工程與應用，55(11)，213-220。
[9]國家發展委員會(2019)。108年持有手機民眾數位機會調查報告。網址：https://ws.ndc.gov.tw/Download.ashx?u=LzAwMS9hZG1pbmlzdHJhdG9yLzEwL2NrZmlsZS9hZjg2Nzg1Ny01YWE0LTRjZTYtODQ3OS00NzVhMWY5NTkyOGMucGRm&n=6ZmE5Lu2OS0xMDjlubTmiYvmqZ%2Fml4%2FmlbjkvY3mqZ%2FmnIPoqr%2Fmn6XloLHlkYot5YWs5ZGK54mILnBkZg%3D%3D&icon=.pdf
[10]張淑青(2009)。目的地意象、滿意度與行為意圖的關係-目的地意象的區隔效果。台灣管理學刊，9(1)，1-22 。
[11]救國團總團部事業處(2010)。日月潭經典八景。
[12]馮淑玲(2020)。大數據在醫療科學上的應用。中山醫學大學檢驗檢驗暨生物技術學系學位醫學論文，台中市。
[13]黃宥綾、施孟隆(2013)。目的地意象、服務品質、滿意度、忠誠度關係之研究-以台東地區鄉村旅遊產業為例。鄉村發展，15，1-20。
[14]鄭欣、田博、李晶晶(2018)。基於YOLO模型的宫頸细胞簇團智能識别方法。液晶與顯示，33(11)，965-971。
[15]蕭丞軒(2020)。以YOLO標註技術擴增資料集增加辨識準確率。中興大學資訊科學與工程學系所學位論文，台中市。

外文文獻：
[1]Banyai, M., Glover, T. D. (2012). Evaluating research methods on travel blogs. Journal of Travel Research, 51 (3), 267-277
[2]Batista, F., Silva, E., Herrera, M. A. M., Rosina, K., Barranco, R. R., Freire, S., & Schiavina, M. (2018). Analysing spatiotemporal patterns of tourism in Europe at high-resolution with conventional and big data sources. Tourism Management, 68, 101-115.
[3]Blackshaw, P., Nazzaro, M. (2016). Consumer-generated media (CGM) 101: Word-of-mouth in the age of the web-fortified consumer. Nielsen BuzzMetrics, New York.
[4]就management: 20 years on and 10 years after the Internet—the state of eTourism research. Tourism Management, 29 (4), 609-623.
[5]Chen, C. F., Tsai, D. (2007). How destination image and evaluative factors affect behavioral intentions?. Tourism Management, 28 (4), 1115-1122.
[6]Chenaghlu, M. A, Derakhshi, M. R., Khasmakhi, N. N., Derakhshi, A. F., Ramezani, M., Nagadeha, Z. J., Farshia, T. R., Moattarad, E. Z., Khadivi, M. R., & Balafarf, M. A. (2021). Chaotic yolo for user intended image encryption and sharing in social media. Information Sciences, 542, 212-227.
[7]Fabricius, S. E., Eriksson, N. (2017). Young-elderly individuals use of social media for travel purpose. Tourism & Management Studies, 13 (1), 27-32.
[8]Fuchs, M., Höpken, W., & Lexhagen, M. (2014). Big data analytics for knowledge generation in tourism destinations–A case from Sweden. Journal of Destination Marketing & Management, 3 (4), 198-209.
[9]Gallarza, M. G., Saura, I. G. & Garcia, H. C. (2002). Destination Image: Towards a Conceptual Framework. Annals of Tourism Research, 29(1), 56-78.
[10]Giannoulakis, S., Tsapatsoulis, N. (2016). Evaluating the descriptive power of Instagram hashtags.J. Innov. Journal of Innovation in Digital Ecosystems, 3 (2), 114-129.
[11]Gretzel, U. (2006). Consumer generated content – trends and implications for branding. e-Review of Tourism Research, 4 (3), 9-11.
[12]Huang, Y. M., Chen, M. Y., & Mo, S. S. (2015). How do we inspire people to contact aboriginal culture with Web2.0 technology?. Computers & Education, 86, 71-83
[13]Hunter, W. C. (2012). Projected destination image: a visual analysis of Seoul. Tourism Geographies, 14 (3), 419-443.
[14]Hunter, W. C. (2016). The social construction of tourism online destination image: A comparative semiotic analysis of the visual representation of Seoul. Tourism Management, 54, 221-229.
[15]Jamtsho, Y., Riyamongkol, P., & Waranusast, R. (2020). Real-time license plate detection for non-helmeted motorcyclist using YOLO. ICT Express, 7(1), 104-109.
[16]Kotler, P., Gertner, D. (2012). Country as brand, product, and beyond: A place marketing and brand management perspective. Journal of Brand Management, 9 (4), 249-261.
[17]Leung, D. H., Lee, A., & Law, R. (2012) Examining hotel managers' acceptance of web 2.0 in website development: A case study of hotels in Hong Kong. Social Media in Travel.Tourism and Hospitality: Theory, Practice and Cases, 53-65.
[18]Li, Y., Lu, Y. J., & Chen, J. (2021). A deep learning approach for real-time rebar counting on the construction site based on YOLOv3 detector. Automation in Construction, 124(3), 103602.
[19]Litvin, S. W., Goldsmith, R. E., & Pan, B. (2008). Electronic word-of-mouth in hospitality and tourism management. Tourism Management, 29 (3), 458-468.
[20]Liu, T., Li, P., Liu, H., Deng, X., Liu, H., & Zhai, F. (2021). Multi-class fish stock statistics technology based on object classification and tracking algorithm. Ecological Informatics, 60, 101240.
[21]Lyu, S. O. (2016). Travel selfies on social media as objectified self-presentation. Tour Manag, 54, 185-195.
[22]MacKay, K. J., Fesenmaier, D. R. (1997). Pictorial element of destination in image formation. Annals of Tourism Research, 24 (3), 537-565.
[23]Mak, H. N. (2017). Online destination image: Comparing national tourism organisation's and tourists' perspectives. Tourism Management, 60, 280-297.
[24]Pan, S. (2011). The role of TV commercial visuals in forming memorable and impressive destination images. Journal of Travel Research, 50 (2), 171-185.
[25]Redmon, J., Divvala, S. K., Girshick, R. B, & Farhadi, A. (2016). You only look once: unified, real-time object detection. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 779-788
[26]Roig, E. M., Rosel, B. F. (2018). Measuring the gap between projected and perceived destination images of Catalonia using compositional analysis. Tourism Management, 68, 236-249.
[27]Sirgy, M. J., Su, C. (2000). Destination image, self-congruity, and travel behavior: Toward an integrative model. Journal of Travel Research, 38 (4), 340-352.
[28]Stepchenkova, S., Zhan, F. (2013). Visual destination images of Peru: Comparative content analysis of DMO and user-generated photography. Tourism Management, 36, 590-601.
[29]Sun, M., Ryan, C., & Pan, S. (2014). Assessing tourists' perceptions and behaviour through photographic and blog analysis: The case of Chinese bloggers and New Zealand holidays. Tourism Management Perspectives, 12, 125-133.
[30]Thompson, K. A. (2006). An eye for the tropics: Tourism, photography, and framing the Caribbean picturesque. Duke University Press Books, Durham, NC.
[31]Tian, Y., Yang, G., Wang, Z., Wang, H., Li, E., & Liang, Z. (2019). Apple detection during different growth stages in orchards using the improved YOLO-V3 model. Computers and Electronics in Agriculture, 157, 417-426.
[32]Ting, H., Wong, W. P. M, Run, E. C., & Lau, S. Y. C. (2015). Beliefs about the use of Instagram: an exploratory study. International Journal of Business and Innovation., 2 (2), 15-31.
[33]Tim, O. (2005). Design Patterns and Business Models for the Next Generation of Software. Web: https://www.oreilly.com/pub/a/web2/archive/what-is-web-20.html.
[34]Volo, S. (2012). Blogs: “Re-inventing” tourism communication. Social Media in Travel. Tourism and Hospitality: Theory, Practice and Cases, 149-163.
[35]Wang, J., Xie, C., Huang, Q., & Morrison, A. M. (2020). Smart tourism destination experiences: The mediating impact of arousal levels. Tourism Management Perspectives,35, 100707.
[36]Waqas, U., Choe, T. (2018). Good Portrait Selection Based on Deep Learning Using Facial Expressions. Journal of Computers, 19(6), 132-139.
[37]Xianga, Z., Gretzel, U. (2010) Role of social media in online travel information search. Tourism Management, 31 (2), 179-188.
[38]Zhang, J., Dong, L. (2021). Image Monitoring and Management of Hot Tourism Destination Based on Data Mining Technology in Big Data Environment. Tourism Management, 60, 280-297.
[39]Zhang, K., Chen, Y., & Li, C. (2019). Discovering the tourists' behaviors and perceptions in a tourism destination by analyzing photos' visual content with a computer deep learning model: The case of Beijing. Tourism Management, 75, 595-608.
[40]Zolkepli, I. A., Hasno, H., Mukhiar, S., & Nadiah, S. (2015). Online Social Network Citizen Engagement on Instagram Crowdsourcing: A Conceptual Framework.Elec. International Journal of Knowledge Management, 13 (4), 283-292